Viewpoint
This article was written by Matt Rees, Chief Technology & Operating Officer at Neos Networks
Matt Rees, Chief Technology & Operating Officer at Neos Networks
No one needs to be reminded of the dominance of AI and its impact on every aspect of our lives. For telecoms, AI is also fueling another, distinct revolution that is fundamentally changing how networks operate. The ever-growing demand for generative AI (GenAI) is driving the rise of ‘edge computing’, with edge data centres being developed and positioned closer to the end-user. By 2030, this market alone is expected to grow by nearly 15% to meet AI’s increasing real-time data processing and low-latency performance demands.
The case for the edge
The case for edge networks is clear. Located close to the areas they serve; edge data centres can significantly reduce latency and boost the performance of applications requiring real-time processing. A shift toward this decentralised approach will help balance loads and maintaining data flows in the event of an outage. Not only do they improve the end-user’s experience, but they also improve the overall resilience of networks for operators.
It’s no secret that AI applications are both data-heavy and compute-intensive, which raises challenges around latency and data storage. With Gartner predicting that GenAI alone will drive a 24% growth in data centres this year, these issues will be exacerbated. However, the edge is set to reduce these pressures on networks.
GenAI requires faster processing time than regular AI applications, so in many cases will require networks to deliver ultra-low-latency. Edge data centres allow enquiries to be stored and processed close to the end-user, promising a faster experience. This isn’t just theoretical either, we are already seeing edge use cases, including predictive maintenance, autonomous vehicles, and immersive experiences – where every millisecond counts.
Sustainability and power consumption of data centres must be considered, particularly given Google’s recent concession that data centre energy consumption significantly contributed to its staggering 48% increase in greenhouse gas emissions. It’s estimated that this year alone, UK businesses will require up to 30% more computing power. However, edge data centres promise to reduce the overall power consumption of the grid due to the wider, distributed network that they create, which spreads the computing burden and power demand more evenly.
The challenges
It isn’t all sunshine and roses, however. The recent news that BT intends to close down 4,600 telephone exchanges, reducing the number dotted around the UK to just 1,000 by the early 2030s, put a spanner in the works for edge data centre operators. These exchanges are vital for the full-fibre rollout across Britain and present opportunities to deliver edge computing services essential for supporting AI. Given BT didn’t yet mention which exchanges it planned to close, the lack of clarity will likely slow down investment decisions and create a race for space within the remaining locations for the hundreds of network operators using these exchanges.
Another issue is a lack of investment. Despite widespread investment in ‘traditional’ data centres such as Google’s $1 billion data centre announced earlier in the year, there’s been less investment in edge data centres. The UK’s ambitious AI strategy must address this; focusing on the size, location, and quality of the underlying infrastructure that will support it. Endeavours like Project Gigabit are crucial steps in the right direction as investment in full-fibre rollout is essential to enabling data centre buildout. However, UK government must also prioritise building out the network edge, not just fibre-to-the-home (FTTH) and the central network.
Data centre buildout = the key to the UK’s AI ambitions
The UK’s goal to become an AI ‘world leader’ will depend on the country’s fixed telecoms infrastructure’s ability to carry significant amounts of data with minimal latency. If our data centres and networks are ill-equipped to deal with the influx of traffic generated by ‘always on’ Large Language Models (LLMs) and other data-hungry applications like IoT and AR/VR, the government’s ambitions could flounder.
A hybrid approach
As AI develops, we anticipate more data centre investment in the North of England. While this will support AI in the region, this investment must be supplemented by edge buildout across the country if the UK is to reach its goal to become an AI superpower. The best approach is a hybrid one, combining strategically placed data centres and adequate PoPs at the network edge in tandem with central data centres. This arrangement will be essential to manage rapid information flow cost-effectively and sustainably while meeting the low-latency needs of AI.
While AI is fueling significant growth in edge computing, its success is reliant on edge data centres being built to strengthen the network. GenAI cannot perform without the low-latency capabilities and real-time processing provided by these facilities. However, they face challenges including minimal investment and the uncertainty created by the BT Openreach exchange closures. The UK government needs to focus on bolstering telecoms infrastructure, including the edge, so that it can handle the increase in data that comes with widescale AI use. Without this, the UK’s fate as an AI superpower hangs in the balance.
Keep up to date with the latest telecoms news by subscribing to the Total Telecom daily newsletter
Also in the news:
Vodafone kick-starts £430 million share buyback scheme
Google Search is an illegal monopoly, US court rules
Deutsche Telekom and Volkswagen leveraging 5G to manage automotive terminal